Modelling and Analysis of Yield by Morpho-Physiological Traits in Wheat under Heat Stress

In Agricultural and Plant Breeding research, via modeling, it is feasible to consider the effective traits in the Yield production (Quantitative or Qualitative) and grant a predictive model for the goal conditions. This experiment was down on ۱۲ wheat genotypes (eight Iranian genotypes and four international genotypes for heat stress research) during two Optimal (Control) and Delayed-Sowing Dates (to create heat stress). According to the results, the Delayed-Sowing Date caused a decrease in the values of most of the Morpho-physiological, and Yield traits compared to control. Based on the percentage of changes compared to the control, the genotypes clustered into three exclusive groups (Tolerant, Sensitive, and Very Sensitive to heat), and Chamran was the high heat tolerant among the Iranian genotypes. Also, the Correlation between most traits was between ۴۱ to ۹۴%. Input values for the modeling by means of the PLS-Regression method consisted of two Predictors (Biomass and Grain Yield) and ۱۹ traits (Responses) and a qualitative variable (Sowing-dates). The Grain Yield variable with a R۲ (۹۳.۶%) and MSE (۷.۱۹۰) confirmed a better fit of the model in contrast to the Biomass Yield variable with R۲ (۹۲.۹%) and MSE (۴۰.۱۷۳). Based on Variable Importance in the Projection (۱.۰ > VIP), the seven traits such as Stomatal Conductance totally eliminated from the two remaining models. The most negative contribution in constructed of model had been associated to the Delayed-Sowing Date (-۳.۴۵۴) for Biomass Yield Predictor, and the No Spikelet in Secondary Spikes × Optimal-Sowing Date (-۲.۶۵۴) for Grain Yield Predictor. The most positive contributions have been associated to No Grains in a Spikelet (۵.۴۵۸) for Grain Yield Predictor and No Fertile Tiller (۱.۶۰۴) for Biomass Yield Predictor. Therefore; they are effective traits in the Predictive Model related to wheat Yield under heat stress.

کلیدواژه ها:

Correlation ، Sowing-Date ، Predictor ، and PLS-Regression